NVIDIA Holoscan SDK v.3.9.0

Class GPUResidentOperator

Base Type

class GPUResidentOperator : public holoscan::Operator

Public Functions

HOLOSCAN_OPERATOR_FORWARD_TEMPLATE () explicit GPUResidentOperator(ArgT &&arg
inline ArgsT std::forward (args)...)
inline GPUResidentOperator()
~GPUResidentOperator()
inline void set_operator_type()
virtual std::shared_ptr<ExecutionContext> initialize_execution_context() override
std::shared_ptr<GPUResidentExecutor> gpu_resident_executor()
std::shared_ptr<cudaStream_t> cuda_stream()

Custom helper functions for GPU-resident Operators below.

Get the CUDA stream for the operator to launch a CUDA workload.

Returns

The CUDA stream for the operator to launch a CUDA workload.

std::shared_ptr<cudaStream_t> data_ready_handler_cuda_stream()

Get the CUDA stream for the operator to launch a data ready handler CUDA workload.

Returns

std::shared_ptr<cudaStream_t>

void *device_memory(const std::string &port_name)

Get the device memory address of an input or output port corresponding to a given port name.

Parameters

port_name – The name of the input or output port.

Returns

The device memory address of the input or output port. Returns nullptr if the port does not exist or is the port does not correspond to a device memory address.

void *data_ready_device_address()

Get the CUDA device pointer for the data_ready signal.

This address can be used in CUDA kernels to signal that data is ready for processing in the GPU-resident execution pipeline. The data ready handler GPU-resident operators can use this address to atomically signal the data is now ready for processing.

Returns

Pointer to the device memory location for the data_ready signal.

void *result_ready_device_address()

Get the CUDA device pointer for the result_ready signal.

This address can be used in CUDA kernels to signal that results from the GPU-resident execution are ready for consumption.

Returns

Pointer to the device memory location for the result_ready signal.

Public Members

ArgsT &&args

Previous Class GPUResidentExecutor
Next Class GPUResourceMonitor
© Copyright 2022-2025, NVIDIA. Last updated on Dec 2, 2025